SMS geo permissions are the rules that decide which destination countries your product will send a one-time password to, and what happens when a request falls outside them. They matter because OTP abuse rarely starts with a stolen account. Someone scripts your signup or login form with numbers from an expensive destination, your server sends a code to each one, and the bill grows while nobody logs in.
Teams usually meet this problem through the invoice. The first reaction is a block list: a few country codes added in a hurry and never reviewed again. Block lists stay one step behind, because traffic moves to the next destination that is still open. An allow list inverts the default — a code is sent only to a destination you have decided to serve, and every other request gets a policy answer instead of a message.
This guide covers the three parts that make geo permissions work: an allowed-countries list for OTP, a mismatch check when a destination is requested from another country, and a step-up path so travellers are not locked out. The examples use Prynt Control, which answers an OTP request with allow, step-up or block plus a reason code, and leaves the sending to your code.
What an OTP request actually tells you
Each code request carries three independent facts. The destination is the country of the phone number you would send to. The origin is the country the request comes from, taken from your edge or your own geo lookup. The actor is the device, network and history behind it — trusted, new or risky.
Geo permissions compare destination and origin, then filter that comparison through the actor. When all three agree, it is a routine send. When the destination is on your list but the origin is a country where you have no users, it is a different decision, even though the number looks fine. The country pair tells you where; the device tells you who, which is what our device fingerprinting signals are for.
Build an allow list, not a block list
Write down the markets you actually serve and put those in the list. In Control the sms use case answers with one of three actions, and each maps to a branch in your code:

allow— send the code.step-up— run a challenge first, then send.block— do not send; return a neutral response.
Keep the list short enough to defend. A useful shape is two tiers: destinations where anyone may request a code, and destinations where only an already-verified account may — an existing customer who is travelling, for example. Everything outside both tiers is declined, and that is a deliberate business decision rather than an accident of configuration.
Flag the mismatch between destination and origin
An allowed destination can still be requested from somewhere else. Treat the mismatch as a signal, not a verdict:
- A customer on holiday requests a code to their home number from abroad. Legitimate.
- A request for a costly destination in a market where you have no users, arriving over a datacenter network. Worth a challenge.
Matching origins keep the send cheap and fast. Mismatched origins get a stronger challenge and a lower send budget for that device and network. One policy at two strengths keeps the allow list from turning into a support queue.
Step up instead of hard-blocking travellers
A hard block on a destination country is the easiest way to break your own login flow. Control returns a suggested step with the step-up answer — captcha, verify_email or card_or_passkey — so the challenge fits the moment:

- Login from a trusted device: verify_email is usually enough.
- Signup with a new number: a captcha in front of the send absorbs scripted requests.
- Bonus or referral flows: card_or_passkey is the stronger gate, because the motive is value, not access.
Two rules keep this honest. Run the challenge before the SMS is sent, never after. And return the same neutral message whether the request was blocked or stepped up, so the response never tells the caller which rule fired.
Set the policy in sms allowCountries
// policy shape, not a library call
const decision = control.decide({
useCase: "sms",
sms: {
allowCountries: ["US", "CA", "GB"],
allowForVerified: ["MX", "ES"],
onMismatch: "step-up"
},
signals: { phoneCountry, requestCountry, deviceTrust }
});
Phone numbers and cards are kept only as keyed hashes, so none of these caps needs the raw number. Geo is one dimension: caps per device, per number and per network sit beside it, because an allowed destination can still be hammered from a single source.
Watch it before you enforce it
Control starts with a 14-day mirror period where every decision is recorded and nothing is blocked. Use it to see how much real traffic falls outside your list, and whether the travellers you were worried about show up as step-ups instead of blocks. Then enforce in stages: block the destinations with no legitimate traffic, step up the rest, and revisit the list when you launch in a new market. Log the reason code with every decision so support can explain a refused code and you can tell which destinations were added for customers and which for scripts.

FAQ
Does an allow list lock out customers who travel?
No, if you separate the cases. Verified accounts can request a code to a destination outside the main list and still meet a step-up on the origin mismatch. The traveller keeps their number; a scripted request to the same country meets the challenge.
Does Prynt send the SMS?
No. Prynt never sends the message and never blocks on its own. Control returns allow, step-up or block with a reason code and a suggested step, and your server decides whether to call your provider.
What happens on day one?
Mirror mode: decisions are recorded for 14 days and nothing is blocked, so you enforce only the rules you have data for. The implementation details are in our docs.
Put the policy in code you control
Geo permissions only pay off when the rule lives in your own request path. Prynt Control gives you the sms use case, the allow list and the caps, with a mirror period to test them — start at pryntid.com, and see how Control sits beside Identify and Replay on the platform page.
Try it free
Prynt is device intelligence with a free plan: 5,000 identifications a month, visitor IDs with confidence and bot detection — no credit card. Upgrade to Pro for VPN, proxy, tampering and behavioral signals.